/
•
•
What it takes to map India
Machine learning and algorithmic muscle aside, there’s plenty of human intervention that goes into mapping India’s roads, buildings and landmarks.
23 November, 2020•14 min
0
23 November, 2020•14 min
0
Getting your Trinity Audio player ready...
Why read this story?
Editor's note: The boyish voice on the other end of the line has a tell-tale Telugu inflection. It belongs to someone who used to work 10 minutes away from Google’s Hyderabad campus. The company that employed him was GlobalLogic, an offshore vendor for a handful of mapping services, including Google Maps, TomTom, and HERE. He asks not to be named, so we’ll call him M for attribution. One had contacted a swathe of former and current GlobalLogic analysts—that’s what mapping operators are called—after learning that not everything in Google Maps is the end result of juiced-up machine learning. M is one of only two who agreed to speak. His day would begin with a “package dump”, which is a high-fidelity satellite image of a specific area. Just like a honeycomb is composed of hundreds of hexagonal cells, each country map is the sum total of hundreds of image packages. Mumbai alone, M shares, has 50-odd image packages. Within GlobalLogic’s Google Maps workforce is a team that maps roads. Its analysts perform the unenviable task of marking every single attribute in a Google …
More in Internet
Internet
Swiggy sits out the quick-commerce race—and risks losing it
Investors eager to ride India’s quick-commerce boom are already losing confidence in Swiggy. A Rs 7,300* crore war chest and little urgency, its restraint is starting to hurt.
